Audio Signature Matching via Spectrogram Noise Filtering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for identifying audio or audiovisual content in real-time are costly, prone to errors due to aspect ratio and frame rate variations, and can be corrupted by extraneous noise, making accurate content recognition challenging.

Innovation Solution

A system that generates and matches audio signatures using a spectrogram analysis, where a device captures audio from a primary source, generates a binary matrix representing peak energy patterns, and sends it to a server for matching against a database, with noise correction mechanisms to ensure accurate identification.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If special hardware in set-top box is used to analyze video, then content identification accuracy is improved, but system cost increases

Engineering Contradiction:
Improvecontent identification accuracyVSAvoidsystem cost
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces complex video analysis hardware with acoustic signal processing. Instead of using specialized set-top box hardware to decode and analyze video frames, the system uses a microphone to capture audio output and processes acoustic signatures through software algorithms, substituting mechanical/video processing with acoustic field analysis.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces audio output and acoustic signature as an intermediary medium between the content source and identification system. Rather than directly analyzing video content, the system captures audio output from speakers, extracts acoustic signatures, and uses these as proxies for content identification, enabling indirect but effective content recognition.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If ancillary codes are added to audio signal, then content identification capability is improved, but robustness to sensor distortions deteriorates

Engineering Contradiction:
Improvecontent identification capabilityVSAvoidrobustness to sensor distortions
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent extracts intrinsic acoustic features and signatures that naturally exist in the audio content rather than adding artificial ancillary codes. By analyzing inherent acoustic properties such as frequency spectra, temporal patterns, and spectral envelopes, the system achieves content identification without introducing vulnerable embedded markers.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms audio signals into different parameter domains (time-frequency representations, spectral features, acoustic descriptors) to create robust signatures. By converting raw audio into transformed parameter spaces through mathematical operations, the system creates identification markers that are inherently resistant to common sensor distortions and environmental variations.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If audio signature is used for content identification, then system cost is reduced, but matching accuracy deteriorates due to extraneous noise

Engineering Contradiction:
Improvesystem costVSAvoidmatching accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent converts the presence of extraneous noise and user-generated audio into beneficial information for identification. By analyzing the interaction between target content audio and background noise, and by using noise robustness as a discriminative feature, the system turns environmental interference into an additional dimension for content verification and identification.

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

Solution Approach 2:

The patent implements feedback mechanisms where the system continuously refines acoustic signature extraction and matching based on observed noise characteristics. By iteratively adjusting extraction parameters and applying noise adaptation algorithms, the system improves matching accuracy in noisy environments through learned feedback from actual operating conditions.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP2954526B1Signature matching of corrupted audio signal
Publication Date: 2019.08.14 ARRIS ENTERPRISES LLC
  • EP2954526B1 patent drawingFigure 1
  • EP2954526B1 patent drawingFigure 2
  • EP2954526B1 patent drawingFigure 3

AI summary

Devices and methods that match audio signatures to programming content stored in a remote database.